Sequential Labelling with Active Learning to Extract Information about Disasters

Başar, M.E.

Sequential Labelling with Active Learning to Extract Information about Disasters

Files

Basar, E._MSc_Thesis_2017.pdf (324.69 KB)

Authors

Başar, M.E.

Issue Date

2017-08-29

Language

en

URI

http://theses.ubn.ru.nl/handle/123456789/5231

Abstract

Learning from past incidents has a great importance for disaster managers. Estimation of the outcomes beforehand can improve preparations for the next incidents. To make this a less labour-intensive task, we aim to automate extracting information from past events. We focus on extracting critical information about flooding events from newspaper articles as our use case. We treat this information extraction task as a sequential labelling task and create an ensemble of two supervised machine learning algorithms, namely Conditional Random Fields and Structured Support Vector Machines, to achieve our goal. However, supervised learning requires manually annotated training data, which is very expensive and time-consuming to obtain. To reduce the need for manual annotation, Active Learning, a human-in-the- loop method, is explored. We obtain improvement on f1-score up to 25% and observe that Active Learning drastically reduces the effort required by annotation.

Supervisor

Farquhar, J.D.R.

Wagemaker, J.

Faculty

Faculteit der Sociale Wetenschappen